提交 f8f71535 编写于 作者: W WenmuZhou

use eps instald of 0.001

上级 a4b0241a
......@@ -25,7 +25,7 @@ class PSELoss(nn.Layer):
ohem_ratio=3,
kernel_sample_mask='pred',
reduction='sum',
**kwargs):
eps=1e-6**kwargs):
"""Implement PSE Loss.
"""
super(PSELoss, self).__init__()
......@@ -34,6 +34,7 @@ class PSELoss(nn.Layer):
self.ohem_ratio = ohem_ratio
self.kernel_sample_mask = kernel_sample_mask
self.reduction = reduction
self.eps = eps
def forward(self, outputs, labels):
predicts = outputs['maps']
......@@ -92,8 +93,8 @@ class PSELoss(nn.Layer):
target = target * mask
a = paddle.sum(input * target, 1)
b = paddle.sum(input * input, 1) + 0.001
c = paddle.sum(target * target, 1) + 0.001
b = paddle.sum(input * input, 1) + self.eps
c = paddle.sum(target * target, 1) + self.eps
d = (2 * a) / (b + c)
return 1 - d
......@@ -104,7 +105,6 @@ class PSELoss(nn.Layer):
.astype('float32')))
if pos_num == 0:
# selected_mask = gt_text.copy() * 0 # may be not good
selected_mask = training_mask
selected_mask = selected_mask.reshape(
[1, selected_mask.shape[0], selected_mask.shape[1]]).astype(
......
Markdown is supported
0% .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑!
想要评论请 注册